News Summaries

China Launches First AI‑Quantum Company ‘Liangzhi Kaiwu’ to Break Computing Limits

At the 2026 Intelligent Quantum Summit in Beijing’s Zhongguancun hub, China unveiled its first dedicated AI‑quantum enterprise, Liangzhi Kaiwu, marking a shift from research labs to a full‑scale industrial platform. The move comes as AI models swell to trillions of parameters and daily token usage tops 140 trillion, pushing classical chips toward an unsustainable power ceiling. Chairman Liu Qingfeng of iFlytek warned that without a new paradigm, the energy needed for future AI could exceed Earth’s current capacity. Liangzhi Kaiwu, a joint venture between iFlytek and Tsinghua University’s atomic‑quantum team, aims to fuse quantum hardware with AI algorithms to accelerate both fields. Highlights include the “Chasing Wind” algorithm that can steer tens of thousands of atoms in milliseconds, and the “Bian Que” error‑correction decoder that tackles quantum noise using AI‑driven diagnostics. Tsinghua researchers recently trapped 10,064 atoms—breaking the 10,000‑atom barrier and surpassing the previous world record. The company will serve three roles: a talent incubator, a frontier research hub, and a launchpad for future quantum‑AI industries. Liu emphasized that success requires scientists from quantum physics, algorithm design, and AI to work as one rope, not as isolated teams. With abundant talent and a strategy that sidesteps the need for ultra‑high‑end chips, China hopes to leapfrog competitors and lay the groundwork for the next generation of general artificial intelligence within the next decade.

Startup Scout AI Raises $100 Million to Build ‘Fury’ – An AI That Could Command Military Drones

Scout AI, a new defense‑tech startup founded by Coby Adcock, just secured a $100 million funding round to develop an artificial‑intelligence system it calls “Fury.” The goal? To give machines the ability to run and even command military equipment, starting with logistics support and eventually moving toward fully autonomous weapons. Chief Technology Officer Collin Otis likens the process to training soldiers: you start with a base level of intelligence—often a large language model (LLM) that’s already been trained on massive data sets—and then teach it the specific skills needed for battlefield decision‑making. While many AI firms shy away from military contracts, Scout says it has partnerships with major cloud providers that supply the underlying pretrained models, though it won’t name the exact providers. The company is already testing drones equipped with vision‑language models that can spot targets, relay images, and even strike without a human in the loop. One envisioned scenario involves a swarm of smaller “munition” drones coordinated by a larger “quarterback” drone that provides extra computing power. The system would scan an area, locate hidden enemy tanks, and engage them, offering a precision alternative to traditional artillery. Scout’s approach raises both excitement about cutting‑edge technology and concerns about the future of autonomous warfare.

AI’s Next Big Shift: From Training to Lightning‑Fast Inference – Nvidia, Google, and China Lead the Charge

The AI industry is moving from a focus on model training to a new era of inference – the real‑time answering of user requests. Analysts say inference will soon make up 73% of all AI workloads, dwarfing the training phase. Nvidia’s CEO Jensen Huang predicts inference demand could be a billion times larger than training, while IDC forecasts this surge will dominate by 2028. Unlike training, which needs massive, short‑term compute power, inference runs continuously and demands ultra‑low latency and cheap per‑token costs. Different applications have very different speed expectations: gamers want a response in under 15 ms, e‑commerce sites around 20 ms, intelligent agents about 50 ms, and chat‑bots up to 100 ms. To meet these needs, hardware makers are rolling out specialized chips. Google unveiled the TPU‑8i, an inference‑first processor that cuts cost‑per‑performance by roughly 80% and speeds up multi‑agent tasks. Nvidia is adding the Groq accelerator to its CUDA ecosystem to offer a high‑speed, low‑throughput option for premium users. Chinese firms are not standing still. Huawei’s Ascend series and Cambricon’s Mengla 590 chip deliver competitive performance with integrated training‑inference instruction sets and robust software stacks, aiming to capture the growing domestic market that now accounts for over 30% of global AI compute. The race is on to build chips that balance speed, efficiency, and price, reshaping how AI services are delivered across the globe.

Ancient Clovis Hunters Chose Tough Quartz—New Research Uncovers the Reason

A fresh look at the stone tools of the Clovis culture is shedding light on a puzzling choice made by some of North America’s earliest big‑game hunters. While most Clovis points are made from high‑quality chert or obsidian—materials that fracture cleanly and are easy to shape—archaeologists have uncovered a handful of points crafted from stubborn quartz crystals, a rock that’s notoriously hard to work with. Lead researcher Dr. Buchanan and his team used a novel scaling method to compare the size, shape, and wear patterns of these quartz tools with those made from more conventional stones. Their findings suggest that the quartz points, despite being more difficult to knap, may have offered specific advantages: greater durability for repeated use, a sharper edge that held longer, or even symbolic value tied to the rarity of the material. The study opens new avenues for exploring how early peoples balanced practicality with cultural preferences. Dr. Buchanan plans to apply the same analytical framework to a broader sample of Clovis artifacts, hoping to determine whether the quartz trend was an isolated experiment or part of a wider, yet understudied, tradition of using “low‑quality” stone for high‑performance tools.

China’s Race to 6G: From Lab Breakthroughs to Everyday Life by 2035

China is fast‑tracking the next generation of mobile networks—6G—toward real‑world use. Experts say each mobile generation takes about a decade, moving through research, standards, product development and finally commercial rollout. In 2018 the country launched its 6G program, building the Purple Mountain Laboratory and appointing You Xiaohu as chief scientist of a dedicated research team. While 5G focused on simply connecting devices, 6G aims to blend communication with sensing, artificial intelligence and even space‑based links, creating an “integrated” network. The Ministry of Industry and Information Technology announced that the first phase of 6G trials is complete, with more than 300 key technologies already in the vault and a second testing phase now under way. The China Academy of Communications and Mobile Communications says the nation is solidifying its 6G vision, hitting early technical milestones and kicking off standards work. Commercial services could appear around 2030, with large‑scale deployment expected by 2035. A distinctive “lay‑the‑eggs‑along‑the‑way” strategy means that core technologies will be piloted in industrial‑internet and digital‑economy settings before the full network launches, helping to iron out challenges and spark new industries. Companies like Purple Star Yu are already testing concepts and building prototypes, proving that 6G can power future devices and business flows.

When Metal Gets Super Thin: 4‑Nanometer Film Defies Expectations

A team of scientists has discovered that a metal film just four nanometers thick—about 25,000 times thinner than a human hair—behaves in a way that surprised even seasoned physicists. Instead of acting like ordinary metal, the ultra‑thin layer develops a built‑in electric field that shifts its work function by more than one electron‑volt, a change that can dramatically affect how electrons move across the material. This unexpected “interfacial polarization” was observed in a sandwich of ruthenium oxide (RuO₂) and titanium dioxide (TiO₂) and could open new doors for designing faster, more efficient electronic components, better catalysts for chemical reactions, and even quantum‑technology devices that rely on precise control of electron flow. The breakthrough was made possible by a collaboration spanning the University of Minnesota, MIT, Texas A&M, the Gwangju Institute of Science and Technology, and other leading labs. Their findings were published in *Nature Communications* and suggest that by carefully engineering strain and thickness at the atomic scale, engineers can fine‑tune material properties in ways previously thought impossible. The work points to a future where ultra‑thin metals become key building blocks for next‑generation technology.

China’s AI‑Powered ‘Intelligent Economy’ Takes Off: Robots, Tokens, and Green Computing Lead the Surge

Artificial intelligence is reshaping China’s economy faster than ever. From chat‑bots that write code to humanoid robots that navigate streets without human help, AI‑driven “intelligent agents” are spawning a wave of one‑person startups and new business models. A key driver is the emerging “token economy,” where every AI request is measured in digital tokens; in March, daily token usage topped 14 trillion – a thousand‑fold jump from early 2024. Companies like Sugon are building ultra‑efficient, cable‑free super‑nodes to boost token output, while telecom giants are shifting from traditional traffic services to token‑based offerings. The government is backing the boom with massive infrastructure projects: ultra‑large AI computing clusters, green‑powered data centers in the Gobi Desert, and a national push to route at least 80 % of new computing power through renewable energy. By the end of March, China’s AI‑focused computing capacity reached 1,882 EFLOPS, and more than 70 “computing corridors” now link major hubs. These policies are already paying off. In the first quarter, value added in AI‑related sectors such as advanced materials, chips, smart equipment, and industrial robots grew between 17 % and 50 %. Analysts predict the core intelligent‑economy market could hit 12.6 trillion yuan by 2030, accounting for a quarter of the digital industry’s revenue, cementing AI as a cornerstone of China’s high‑quality growth.

Breakthrough Nanotube Fibers Near Copper’s Conductivity – A New Era for Lightweight Wiring

A team of researchers led by de Isidro‑Gómez has unveiled a new class of carbon‑nanotube (CNT) fibers that dramatically boost electrical performance, bringing them within striking distance of copper’s famed conductivity. By intercalating metal ions between tightly packed nanotubes, the scientists created fibers that not only conduct electricity with a specific conductivity rivaling copper but also weigh far less, offering a high strength‑to‑weight ratio ideal for aerospace, automotive, and wearable electronics. The study, published in *Science* (2026), details how the intercalation process aligns the nanotubes, reduces electron scattering, and enhances charge transport pathways. Tests showed the CNT fibers maintained stable conductivity under bending and stretching, outperforming traditional copper wires in durability tests. If scaled up, this technology could slash the weight of power‑distribution systems, improve energy efficiency, and reduce reliance on copper mining. The breakthrough points to a future where ultra‑light, highly conductive cables replace heavy metal conductors, unlocking new design possibilities for next‑generation devices and infrastructure.

Hidden Heat Deep Below: Why Sea Levels Are Rising Faster Than Expected

A new study has uncovered the missing piece in the puzzle of rising sea levels. Scientists led by Anny Cazenave, an emeritus researcher at France’s LEGOS laboratory, discovered that the deep ocean—below 2,000 meters—has been quietly heating up and expanding. This hidden warmth adds enough volume to the world’s oceans to explain a long‑standing “anomaly” where observed sea‑level rise didn’t match predictions based only on surface warming. By including the deep‑ocean expansion in their calculations, the researchers were able to balance the global sea‑level budget almost perfectly for the period since 2016. The findings highlight that the ocean’s interior, far from the reach of satellites and buoys, plays a crucial role in climate dynamics. As the deep sea continues to absorb heat, it not only contributes to higher sea levels but also influences long‑term climate patterns, including the stability of ice sheets in Antarctica. This research underscores the importance of monitoring the entire ocean column—not just the surface—to better predict future sea‑level changes and plan for coastal resilience.